Raquel Garcia:
|
Evidence for shifts in the phenologies and distributions of species over recent decades has often been attributed to climate change. The prospect of greater and faster changes in climate during the 21st century has spurred a stream of studies anticipating future biodiversity impacts. Yet, uncertainty is inherent to both projected climate changes and their effects on biodiversity, and needs to be understood before projections can be used. This thesis seeks to elucidate some of the uncertainties clouding assessments of biodiversity impacts from climate change, and explores ways to address them. While the focus is mostly on sub- Saharan African vertebrates, the methodological advances and conclusions presented are farreaching and have wider relevance.
Throughout the chapters in this thesis, projections under changing climates for sub- Saharan African vertebrates, based on bioclimatic envelope models, are shown to be affected by multiple uncertainties. Different model algorithms produce different outputs, as do alternative future climate models and scenarios of future emissions of greenhouse gases. Another uncertainty arises due to omission of species with small sample sizes, which are difficult to model. The effect of such bias against narrow-ranging species is often overlooked in assessments of biodiversity impacts, but our results for sub-Saharan African amphibians show that it trickles down to conservation strategies. Finally, assumptions about the climatic tolerance of species, their dispersal ability, and other characteristics are also shown to alter model projections for sub-Saharan African amphibians.
Despite numerous calls to address the uncertainty challenge, appropriate treatment of uncertainty has yet to be formalised in assessments of biodiversity impacts under climate change. The chapters in this thesis highlight the need to both integrate uncertainties in assessments, and reduce or circumvent them where possible. Integration of uncertainties is illustrated in two examples. The first uses ensembles of bioclimatic envelope models for sub- Saharan African vertebrates, built with alternative climate data and model algorithms. Ensemble forecasting provides a means for exploring the breadth and spatial variation of uncertainties, and for building consensus among projections. Several consensus methodologies are compared here, including a newly proposed methodology that preserves information about the variability of projections in the ensemble. The second example examines model outputs for sub-Saharan African amphibians in the light of species' vulnerability to climate change. An analytical framework is developed for distinguishing between different climatic threats and opportunities revealed by the bioclimatic envelope models, and analysing how they each are altered by the consideration of specific response-mediating traits.
Efforts to reduce uncertainties in biodiversity impact assessments are equally important. However, many sources of uncertainty cannot easily be reduced, not least the omission of species that are narrow-ranging, poorly known, or even unknown to science. This uncertainty can instead be circumvented through the use of alternative approaches to assessing impacts. This thesis discusses one candidate approach that is independent of species' data: the use of climate change metrics. By describing the exposure of regions to multiple changes in the magnitude, timing, position, or availability of climatic conditions, metrics can provide inferences about the potential threats and opportunities for the biodiversity in those regions. The diversity of existing metrics is reviewed here, and the picture that emerges is one of multifaceted changes in climate, with unequal spatial patterns around the world. To help interpret the diversity of climate change metrics, a conceptual framework is proposed for using them in biodiversity impact assessments. Early testing of this framework, by comparing inferences from metrics and from the bioclimatic envelope models for sub-Saharan African amphibians, suggests that climate change metrics might be a useful addition to the biodiversity impact assessment toolbox.
The uncertainties discussed in this thesis, and many others not covered here, impair the conservation of biodiversity under changing climates in Africa and elsewhere. Explicitly addressing all uncertainties of projected impacts appears overwhelming. Yet, if model projections are to be useful for conservation planners, they must be as transparent as possible by including an honest description of their level of confidence given the current knowledge.